Neuropsychiatric outcomes of ketamine and midazolam coadministration following vascular surgery: a multicenter database study

Introduction

Vascular surgery carries a high opioid burden, with one retrospective study finding that opioids were prescribed for 78% of common vascular procedures while 70% of patients reported opioid use (1). Given ongoing national efforts to combat opioid-related abuse and overdose, opioid-sparing strategies such as ketamine are particularly appealing among this surgical population (2). Although the use of ketamine has the potential advantage of providing an anesthetic regimen that is opioid-sparing, recent years have seen growing attention paid to the potential neuropsychiatric sequelae associated with perioperative ketamine use (3, 4). One study demonstrated that ketamine use during routine orthopedic surgery was associated with increased 30-day risk of new-onset psychiatric comorbidities including depression, anxiety, and psychosis (3).

Patients with peripheral vascular disease needing vascular surgery commonly exhibit comorbidities that heighten their susceptibility to ketamine's psychomimetic effects, including hallucinations, agitation, delirium, and confusion (1, 5–7). This is because the pathophysiology that underlies the complex mechanism for developing peripheral vascular disease is often shared with both cardiovascular and cerebrovascular disease and both diseases are independent risk factors for postoperative neurocognitive dysfunction (8–10).

Thus, the vascular surgery patient population most likely to benefit from ketamine may also be most vulnerable to its psychiatric adverse effects. Adding the use of midazolam when ketamine is administered is a common clinical practice to reduce ketamine-associated emergence events (11). Reducing emergence events may be beneficial because it has been demonstrated that emergence delirium is an independent risk factor for 30-day neurocognitive sequelae (12, 13). Since neurocognitive impairment and changes in psychiatric states are complex processes that are often intertwined, we raise the question of whether the use of midazolam in addition to ketamine would reduce the risk of developing postoperative psychiatric sequelae compared to an anesthetic regimen that included ketamine but without midazolam administration. This represents a clinically relevant knowledge gap given this patient population's vulnerability (6, 7). In the present study, we used a national multicenter database to compare risks of 30-day psychiatric outcomes following vascular surgery between patients receiving ketamine alone and those receiving ketamine and midazolam as part of the anesthetic regimen.

Methods

Data extraction

This retrospective cohort study utilized the TriNetX Research Network (TriNetX, LLC, Cambridge, MA, USA), a real-time database providing access to deidentified electronic health records from over 100 million patients from 70 healthcare organizations (HCOs) within the US Collaborative Network. Palchuk et al. provides further details about the TriNetX database (14). An HCO is defined by TriNetX as a large academic institution encompassing its main hospitals, satellite hospitals, and outpatient clinics. The data in this study were collected and analyzed on 27 July 2026 from TriNetX. This study was conducted using fully deidentified data with no direct interaction with human subjects. In accordance with federal regulations and institutional guidelines, this study was determined to be exempt from Institutional Review Board (IRB) review and did not require informed consent. TriNetX complies with the Health Insurance Portability and Accountability Act (HIPAA) and general data protection regulation. All research adhered to the tenets of the Declaration of Helsinki and the Strengthening the Reporting of Observational Studies in Epidemiology guidelines (15).

Cohort selection

Patients with diagnoses and medications of interest in our study were identified using the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) and RxNorm codes (Supplementary Table 1). Patients were included if they were over 18 years of age and underwent vascular surgery procedures between July 2006 and July 2026. Patients in the 14- and 30-day outcome groups required a follow-up visit within 14 and 30 days postoperatively, respectively. Procedures were identified by the current procedural terminology (CPT) codes listed in the Dartmouth Atlas of Vascular Health Care (16) (Supplementary Table 2). This atlas has been widely used in previous studies to identify CPT codes for a broad range of vascular interventions (Supplementary Table 2). Procedures that are typically performed with minimal to moderate sedation without involving anesthesia, such as varicose vein excision, transcatheter therapy infusion for thrombolysis, and exchange of arterial catheter during thrombolysis, were not included in this study. The exclusion of these procedural types was based on the rationale that anesthetic regimens between non-anesthesia providers and anesthesia providers are inherently different.

Exclusion criteria

Patients were excluded if they had a documented history of mental, behavioral, or neurodevelopmental disorders; central pain syndrome; chronic pain syndrome; chronic pain not elsewhere classified; ischemic heart disease; heart failure; or cerebrovascular disease (Supplementary Table 1). These conditions were selected for exclusion because they are known risk factors for postoperative delirium and/or postoperative cognitive dysfunction. Both postoperative delirium and/or cognitive dysfunction have significant clinical implications for the development of psychiatric-related sequelae. In addition, patients were excluded if they had experienced postoperative major adverse cardiac events (MACE) or were deceased within 31 days of surgery. MACE were defined as including ischemic heart disease, heart failure, or cerebrovascular disease. Excluding patients who had postoperative MACE was based on the same rationale that the development of cardiovascular disease is clinically relevant in the development of cognitive/psychiatric-related sequelae.

Anesthetic regimen definition

Information on dosing of medications is not available on the TriNetX platform. Similarly, information on types of anesthesia such as general anesthesia vs. monitored anesthesia care is not available. As such, information on the use of volatile agents, typically indicative of general anesthesia, is not available. The cohorts were specified based on a binary definition of whether the patients in each group received ketamine or not, as follows: (1) a ketamine alone group, (2) a ketamine and midazolam group, and (3) a control group that received neither ketamine nor midazolam. The use of other common anesthetic agents such as propofol, opioids, and less commonly, dexmedetomidine, in each cohort could not be extracted after propensity score matching (PSM) was performed for each comparison due to the inherent limitations of the TriNetX platform.

Primary outcome measures

The primary outcome was defined as any new psychiatric diagnosis made within 30 days postoperatively. Only patients without preexisting psychiatric diagnoses and with documented follow-up within 30 days were included in the primary analysis. Four distinct categories of psychiatric outcomes were defined based on ICD-10 diagnostic codes. Psychotic sequelae were defined as any documented schizophrenia, schizotypal, delusional, and other non-mood psychotic disorders. Anxiety-related sequelae were defined as anxiety, dissociative, stress-related, somatoform, and other non-psychotic mental disorders. Mood-related sequelae were defined as mood disorders. Behavior-related sequelae were defined as either behavioral syndromes associated with physiological disturbances and physical factors or mental and behavioral disorders due to psychoactive substance use. Specific ICD-10 codes can be found in Supplementary Table 1. In addition, an overall composite outcome encompassing all four domains was evaluated.

Secondary exploratory outcome

In a similar manner, we defined the secondary exploratory outcome as any new psychiatric diagnosis made between 1 and 14 days postoperatively. Only patients without preexisting psychiatric diagnoses and with documented follow-up between 1 and 14 days were included in the secondary analysis. The same set of psychiatric domains and overall composite outcomes used in the primary outcomes measure were used in the secondary exploratory analysis.

Statistical analysis

All statistical analyses were performed using the TriNetX analytics platform in March 2025. PSM was used to minimize the effect of confounding variables. TriNetX uses logistic regression to generate propensity scores based on selected covariates such as age, sex, race, ethnicity, and education/socioeconomic status (Supplementary Table 1). Patients were matched using a greedy nearest neighbor matching algorithm without replacement based on propensity scores to create a 1:1 ratio. Furthermore, patients were matched using a default caliper width of 0.1 and the cohorts were randomly shuffled prior to matching. The covariate balance was examined using standard mean differences (SMDs). An SMD of less than 0.1 was considered well-matched.

Outcome comparisons were conducted using the “compare outcomes” function in TriNetX. Two cohorts were entered into this function at a time and analyzed, as follows: ketamine vs. control, ketamine and midazolam vs. control, and ketamine vs. ketamine and midazolam. Each psychiatric diagnosis group (e.g., mood, anxiety) was set as the outcome. Thus, each exposure and outcome variable served as a binary variable for analysis. TriNetX utilizes 2 × 2 contingency tables to calculate the odds ratios (ORs), risk ratios, and risk differences and uses a Z-test to generate p-values and 95% confidence intervals. Statistical significance was defined as p < 0.05.

Results

Ketamine vs. control

One day postoperatively

Before PSM, the cohort that received ketamine consisted of 1,821 patients, while the control cohort comprised 86,588 patients (Table 1). After PSM, the ketamine and control cohorts were balanced to 1,821 patients each (Table 1). Although an SMD threshold of <0.1 was prespecified to define acceptable balance, all measured covariates achieved excellent balance with an SMD < 0.01. There was a notable prematch difference in Hispanic ethnicity between the groups (8.7% in the ketamine group vs. 16.7% in the control group; SMD=0.24), which was successfully resolved following matching (SMD <0.002). Patients who received ketamine only on the day of their surgery were found to have an OR of 2.33 [95% confidence interval (CI): 1.89–2.88, p < 0.0001] for the composite new psychiatric diagnosis outcome, an OR of 1.77 (95% CI: 1.19–2.64, p = 0.0044) for new mood-related diagnoses, an OR of 2.29 (95% CI: 1.61–3.26, p < 0.0001) for new anxiety-related diagnoses, and an OR of 2.73 (95% CI: 2.12–3.52, p < 0.0001) for new behavior-related diagnoses within 1 day of surgery compared to the control cohort. There was an insufficient number of events in the psychotic outcome domain in both groups for statistical analysis (Table 2).

Table 1. Demographic characteristics of the ketamine only and control cohorts (1 day) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Table 2. Absolute event rates, odds ratios, 95% confidence intervals, and p-values for each cohort 1 day postoperatively.

OR, odds ratio; CI, confidence interval; K, ketamine; C, control; KV, ketamine and midazolam.

Psychotic outcomes could not be ascertained for the K vs. C and K vs. KV analyses due to inadequate outcome event sizes.

Bolded values indicate statistical significance (P < 0.05).

Fourteen days postoperatively

Before PSM, the cohort that received ketamine consisted of 1,485 patients, while the control cohort comprised 61,145 patients (Table 3). After PSM, the ketamine and control cohorts were balanced to 1,485 patients each (Table 3). All measured covariates achieved excellent balance following matching (all SMD ≤0.01). There was a notable prematch difference in Hispanic ethnicity between the groups (9.0% in the ketamine group vs. 17.3% in the control group; SMD=0.25), which was successfully resolved following matching (SMD <0.001). Patients who received ketamine only on the day of their surgery were found to have an OR of 2.22 (95% CI: 1.79–2.75, p < 0.0001) for composite new psychiatric diagnoses, an OR of 1.71 (95% CI: 1.13–2.58, p = 0.0106) for new mood- related diagnoses, an OR of 2.11 (95% CI: 1.48–3.02, p < 0.0001) for new anxiety-related diagnoses, and an OR of 2.65 (95% CI: 2.04–3.45, p < 0.0001) for new behavior-related diagnoses within 14 days of surgery compared to the control cohort. There was an insufficient number of events in the psychotic outcome domain in both groups for statistical analysis (Table 4).

Table 3. Demographic characteristics of the ketamine only and control cohorts (14 days) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Table 4. Absolute event rates, odds ratios, 95% confidence intervals, and p-values for each cohort 14 days postoperatively.

OR, odds ratio; CI, confidence interval; K, ketamine; C, control; KV, ketamine and midazolam.

Psychotic outcomes could not be ascertained for the K vs. C and K vs. KV analyses due to inadequate outcome event sizes.

Bolded values indicate statistical significance (P < 0.05).

Thirty days postoperatively

Before PSM, the cohort that received ketamine only consisted of 1,424 patients, while the cohort receiving no ketamine or midazolam included 58,564 patients (Table 5). After PSM, the ketamine and control cohorts were balanced to 1,424 patients each (Table 5). All measured covariates achieved excellent balance following matching (all SMD <0.01). There was a notable prematch difference in Hispanic ethnicity between the groups (9.1% in the ketamine group vs. 18.1% in the control group; SMD=0.26), which was successfully resolved following matching (SMD=0.002). Patients who received ketamine only on the day of their surgery were found to have an OR of 2.25 (95% CI: 1.81–2.79, p < 0.0001) for composite new psychiatric diagnoses, an OR of 1.69 (95% CI: 1.12–2.54, p = 0.0113) for new mood-related diagnoses, an OR of 2.21 (95% CI: 1.57–3.12, p < 0.0001) for new anxiety-related diagnoses, and an OR of 2.79 (95% CI: 2.13–3.65, p < 0.0001) for new behavior-related diagnoses within 30 days of surgery compared to patients receiving no ketamine or midazolam. There was an insufficient number of events in the psychotic outcome domain in both groups for statistical analysis (Table 6).

Table 5. Demographic characteristics of the ketamine only and control cohorts (30 days) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Table 6. Absolute event rates, odds ratios, 95% confidence intervals, and p-values for each cohort 30 days postoperatively.

OR, odds ratio; CI, confidence interval; K, ketamine; C, control; KV, ketamine with midazolam.

Psychotic outcomes could not be ascertained for K vs. C and K vs. KV analysis due to inadequate outcome event size for analysis.

Bolded values indicate statistical significance (P < 0.05).

Ketamine and midazolam vs. control

One day postoperatively

Before PSM, the cohort that received ketamine and midazolam consisted of 8,318 patients, while the control group consisted of 85,406 patients before PSM (Table 7). After PSM, both groups were balanced to 8,318 patients each (Table 7). The most significant prematching difference was in age, with the patients who received ketamine and midazolam 7.4 years younger on average than controls (52.6 ± 17.2 vs. 60.0 ± 18.2 years; SMD=0.41). Ethnicity also differed substantially, with 78.4% of the ketamine and midazolam cohort documented as not Hispanic or Latino vs. 59.3% of controls (SMD=0.42). Both imbalances were resolved following matching, and all measured covariates achieved excellent balance (SMD <0.05). The patients in the ketamine and midazolam group were found to have an OR of 1.83 (CI: 1.67–2.00, p < 0.0001) for composite new psychiatric diagnoses, an OR of 1.72 (CI:1.44–2.05, p < 0.0001) for new mood-related diagnoses, an OR of 2.34 (CI: 1.99–2.74, p < 0.0001) for new anxiety-related diagnoses, and an OR of 1.67 (CI: 1.50–1.86, p < 0.0001) for new behavior-related diagnoses compared to the control group (Table 2). For the psychotic outcome domain, the difference between the groups did not meet statistical significance (OR 1.31, CI: 0.76–2.25, p = 0.3355) (Table 2).

Table 7. Demographic characteristics of ketamine with midazolam vs. control cohort (1 day) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Fourteen days postoperatively

Before PSM, the cohort that received ketamine and midazolam consisted of 7,393 patients, while the control group comprised 61,627 patients (Table 8). After PSM, both groups were balanced to 7,392 patients each (Table 8). Similar to the 1-day analysis, the ketamine and midazolam cohort was much younger than controls before matching (52.7 ± 17.2 vs. 60.5 ± 18.2 years; SMD=0.45), with a similar difference in ethnicity (80.8% vs. 58.7% documented as not Hispanic or Latino; SMD=0.50). Both were resolved following matching, with all measured covariates achieving excellent balance (all SMD ≤0.05). The patients in the ketamine and midazolam group were found to have an OR of 2.20 (CI: 1.99–2.41, p < 0.0001) for composite new psychiatric diagnoses, an OR of 1.77 (CI:1.49–2.11, p < 0.0001) for new mood-related diagnoses, an OR of 2.56 (CI: 2.19–2.98, p < 0.0001) for new anxiety-related diagnoses, and an OR of 2.09 (CI: 1.86–2.34, p < 0.0001) for new behavior-related diagnoses compared to the control group. For the psychotic outcome domain, the difference between the groups did not meet statistical significance (OR 1.28, CI: 0.76–2.16, p = 0.3529) (Table 4).

Table 8. Demographic characteristics of the ketamine and midazolam and control cohorts (14 days) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Thirty days postoperatively

Before PSM, the cohort that received ketamine and midazolam consisted of 6,649 patients, while the control group comprised 58,564 patients (Table 9). After PSM, both groups were balanced to 6,644 patients each (Table 9). The ketamine and midazolam cohort was again significantly younger before matching (52.4 ± 17.1 vs. 60.0 ± 17.9 years; SMD=0.44), with an ethnicity difference (80.4% vs. 61.9% documented as not Hispanic or Latino; SMD=0.42). Both were resolved following matching, with all measured covariates achieving excellent balance (SMD ≤0.05). The patients in the ketamine and midazolam group were found to have an OR of 2.00 (CI: 1.82–2.20, p < 0.0001) for composite new psychiatric diagnoses, an OR of 1.53 (CI:1.29–1.82, p < 0.0001) for new mood-related diagnoses, an OR of 2.63 (CI: 2.26–3.06, p < 0.0001) for new anxiety-related diagnoses, and an OR of 1.85 (CI: 1.65–2.07, p < 0.0001) for new behavior-related diagnoses compared to the control group. For the psychotic outcome domain, the difference between the groups did not meet statistical significance (OR 1.29, CI: 0.79–2.11, p = 0.3161) (Table 6).

Table 9. Demographic characteristics of the ketamine and midazolam and control cohorts (30 days) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Ketamine vs. ketamine and midazolam

One day postoperatively

Before PSM, the ketamine only group comprised 1,821 patients, while the ketamine and midazolam group comprised 8,804 patients. After PSM, both cohorts were equalized to 1,786 patients (Table 10). The ketamine cohort was older than the ketamine and midazolam cohort (62.7 ± 18.9 vs. 52.9 ± 17.2 years; SMD=0.54). This was resolved following matching (SMD=0.02). All measured covariates achieved the prespecified SMD threshold (SMD <0.1). The difference between the groups did not meet statistical significance for composite new psychiatric diagnosis (OR 1.11, 95% CI: 0.93–1.33, p = 0.2382), for new mood-related diagnosis (OR 1.11, 95% CI: 0.79–1.59, p = 0.5302), or for new anxiety-related diagnosis (OR 1.00, 95% CI: 0.75–1.33, p = 0.9988). However, the ketamine only cohort was found to have an OR of 1.34 (95% CI: 1.08–1.65, p = 0.0066) for behavior-related diagnoses compared to the ketamine and midazolam group. There was an insufficient number of events in the new psychotic outcome domain in both groups for statistical analysis (Table 2).

Table 10. Demographic characteristics of the ketamine only and ketamine and midazolam cohorts (1 day) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Fourteen days postoperatively

Before PSM, the ketamine only group comprised 1,457 patients, while the ketamine and midazolam group comprised 7,393 patients. After PSM, both cohorts were equalized to 1,420 patients (Table 11). Consistent with the 1-day comparison, the ketamine only cohort was older before matching (63.2 ± 18.8 vs. 52.7 ± 17.2 years; SMD=0.58). This age difference was resolved following matching (SMD=0.01). All measured covariates achieved acceptable balance (SMD <0.1). The ketamine only cohort was found to have an OR of 1.29 (95% CI: 1.06–1.57, p = 0.0113) for composite new psychiatric diagnoses and an OR of 1.54 (95% CI: 1.22–1.95, p = 0.0002) for new behavior-related diagnoses compared to the ketamine and midazolam group (Table 4). New mood-related diagnoses and anxiety-related diagnoses did not differ significantly between groups (OR 1.40, 95% CI: 0.95–2.08, p = 0.0909 and OR 1.00, 95% CI: 0.74–1.35, p = 0.9980, respectively). There was an insufficient number of events in the new psychotic outcome domain in both groups for statistical analysis (Table 4).

Table 11. Demographic characteristics of the ketamine only and ketamine and midazolam cohorts (14 days) before and after propensity score matching.

SMD, standardized mean difference; SD, standard deviation.

Thirty days postoperatively

Before PSM, the ketamine only group consisted of 1,424 patients, while the ketamine and midazolam group consisted of 6,649 patients. These cohorts were equalized to 1,359 patients each after PSM (Table 12). The ketamine only cohort was older before matching (62.4 ± 18.8 vs. 52.2 ± 17.2 years; SMD=0.56). This age difference was resolved following matching (SMD=0.01). All measured covariates achieved acceptable balance (SMD <0.1). The ketamine only cohort was found to have an OR of 1.22 (95% CI: 1.01–1.48, p = 0.0438) for composite new psychiatric diagnoses and an OR of 1.44 (95% CI: 1.14–1.81, p = 0.0016) for behavior-related diagnoses compared to the ketamine and midazolam group. For the mood-related and anxiety-related diagnosis domains, the differences between the groups did not achieve statistical significance (OR 1.23, 95% CI: 0.84–1.79, p = 0.2905 and OR 1.06, 95% CI: 0.79–1.40, p = 0.7152, respectively). There was an insufficient number of events in the new psychotic outcome domain in both groups for statistical analysis (Table 6).

Table 12. Demographic characteristics of the ketamine only and ketamine and midazolam cohorts (30 days) before and after propensity score matching.

Discussion

In this large retrospective cohort study of patients who underwent vascular surgery, we found that patients who received ketamine only, compared to ketamine and midazolam coadministration, had increased odds for a new behavior-related diagnosis within 30 days postoperatively. Although there was a statistically significant difference between the groups in the odds of a composite new psychiatric diagnosis, predefined to encompass all four domains (psychotic, mood, anxiety, and behavioral), it is likely that the new behavior-related diagnosis domain was the main driving factor explaining this difference between the groups. This primary finding was supported by our secondary exploratory findings that the patients who received ketamine only also exhibited increased odds of a new behavior-related diagnosis at 1 and 14 days postoperatively.

Together, these findings suggest that a new behavior-related phenomenon related to ketamine use may manifest as early as 1 day postoperatively and may extend well into the 30-day postoperative window. In addition, our analysis showed that coadministration of ketamine and midazolam compared to ketamine only was associated with reduced odds of development of new behavior-related sequelae at all three postoperative time points. However, coadministration of ketamine and midazolam, compared to the non-exposed group, was associated with higher odds of developing three out of the four domains of psychiatric sequelae (mood-, anxiety-, and behavior-related) and higher odds across all three time points (1, 14, and 30 days), in a similar manner as observed in the comparison between the ketamine alone and control groups. These findings suggest that the role of the midazolam-ketamine combination and its impact on long-term psychiatric sequelae in an adult surgical cohort is complex. Further prospective and randomized controlled studies are necessary to elucidate the role of the midazolam-ketamine combination on 30-day psychiatric outcomes in surgical cohorts.

To our knowledge, our study is the first to explore the role of midazolam-ketamine coadministration and 30-day psychiatric/behavioral outcomes in adult surgical patients. Indeed, previous studies have largely investigated the topic of midazolam-ketamine coadministration and its implications for cognitive and psychiatric outcomes in the immediate postprocedural or postsurgical period. For example, Sener et al., in their prospective and double-blinded study conducted in an emergency department, demonstrated a reduced incidence of recovery agitation when midazolam was coadministered with ketamine in procedural sedation compared with ketamine alone (17). In another interesting study by Suzuki et al., midazolam was found to attenuate ketamine-induced altered perceptions and thought processes (18). In more comparative studies on adult surgical cohorts, ketamine and midazolam premedication was shown to be effective in reducing acute emergence phenomena and other psychiatric sequelae in other short urological procedures (11, 19).

As mentioned previously in the Introduction, the rationale for the current study was based on the current evidence that the development of emergence delirium is strongly associated with the development of 30-day neurocognitive sequelae (12, 13), and neurocognitive impairment is one of the strongest independent risk factors for long-term psychiatric morbidity in the postoperative period (20). Therefore, strategies to mitigate the development of emergence delirium may also reduce the risk of developing long-term psychiatric sequelae.

Clinically, the coadministration of midazolam and ketamine is commonly practiced, with the intended effect of reducing the risk of ketamine-related neurocognitive and/or psychiatric changes in the immediate postoperative period. Besides some supportive evidence for this practice in the literature (11, 19), the rationale for combining midazolam and ketamine is also based on pharmacological principles: midazolam's GABA-A agonism is intended to counteract the psychotomimetic side effects of ketamine's NMDA receptor antagonism (21).

However, whether midazolam has a protective effect and can mitigate ketamine-induced emergence delirium postoperatively is still unclear (11). Moreover, studies investigating long-term psychiatric effects of perioperative midazolam have had mixed findings (3, 22–24). Although our study found that ketamine-midazolam coadministration was associated with reduced odds of developing behavior-related psychiatric sequelae compared to ketamine-only administration, this result should be interpreted with caution due to several study limitations.

Study limitations

As the study was performed retrospectively, there is an inherent risk for residual confounding despite propensity score matching. While we excluded cohorts with pre-existing conditions that are known to contribute to postoperative cognitive dysfunction, there are several important variables we could not control. For example, information on dosing and timing of both ketamine and midazolam was unavailable. It is likely that both midazolam's potential protective effect and ketamine's psychomimetic effect are dose-dependent. Moreover, midazolam is known to lead to anterograde amnesia and is typically administered before surgery. However, it is unknown whether the sequence and timing of midazolam administration in relation to ketamine administration has a relevant impact on our outcomes of interest.

Other examples of relevant variables that we could not account for due to the inherent limitations of the TriNetX database include surgical duration, anesthetic technique (e.g., general vs. regional), and specific anesthetic agent exposures, such as dexmedetomidine, propofol-based total intravenous anesthesia, narcotics, and volatile anesthetics. This information has relevant implications since the control cohort likely included patients who were exposed to anesthetic agents that may also have neuropsychiatric risk profiles. All these factors may influence postoperative psychiatric outcomes.

Frailty is another important variable that has an important implication for postoperative cognitive/psychiatric outcomes. Multiple studies have shown strong associations between frailty and the development of neuropsychiatric disorders postoperatively (25–27). We could not investigate this important risk factor because the frailty index is not directly included in TriNetX. Given the complexity of this index, which includes not only multiple comorbid diseases but also functional status and phenotypic metrics, it would not be possible to calculate this index accurately from the data that is available in TriNetX. However, we note that frailty is known to increase in incidence as the age of the study cohort increases. One meta-analysis reported a prevalence of frailty among people aged 50–59 of 11%, with prevalences of 16%, 20%, and 31% among those aged 60–69, 70–79, and 80–89, respectively(28). The mean age of our patient cohorts ranged from 50 to 65. Thus, our study cohort is relatively younger in mean age than the older cohort in which the incidence of frailty is reported to be 20%–30%. Nonetheless, we understand that frailty is an important risk factor that should be incorporated in future studies.

In our secondary exploratory analysis, we could not investigate the outcomes in the immediate postsurgical timeframe that would be indicative of emergence delirium. This would have supported the main rationale of this study if increased odds of association were found between the study cohorts and outcomes of interest.

Our inclusion criteria required documented follow-up within 30 days, and the proportion of patients excluded due to absent follow-up could not be determined. Differential follow-up between groups could introduce bias toward a certain regimen if one cohort was more likely to receive postoperative clinical evaluation. Finally, the study focused primarily on patients who underwent vascular surgery, which limits generalizability to other surgical candidates.

Conclusion

In this retrospective analysis, coadministration of ketamine and midazolam was associated with reduced odds of developing a behavior-related psychiatric diagnosis within 30 days after vascular surgery compared with ketamine alone in adult cohorts. Due to the major limitations of this study, this result should be interpreted with caution. Indeed, the result should not be interpreted as evidence that midazolam has a protective effect in the prevention of ketamine-associated psychomimetic reactions. Rather, the result of this study serves as an exploratory basis for future research. Future research involving prospective randomized controlled trials is necessary to better define the impact of midazolam-ketamine coadministration on long-term psychiatric sequelae.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fanes.2026.1888859/full#supplementary-material

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Ansgar Brambrink, Columbia University, United States

Reviewed by: Sebastian Zinn, Goethe University Frankfurt, Germany

Ruediger Noppens, University of Ottawa, Canada